Associations between mixtures of urinary phthalate metabolites with gestational age at delivery: a time to event analysis using summative phthalate risk scores

Abstract

[...]using subject-specific intercepts is a more general approach as random slopes or other features may be incorporated into the stage 1 linear mixed model (LMM). [...]the LMM framework better addresses subjects with differing numbers of visits, because BLUPs are shrinkage estimates of subject-specific averages relative to the population average. First-step models adjusted for time-varying specific gravity and all second-step models adjusted for race, education, maternal age, and health insurance status. MECPP (HR: 1.11, 95% CI: 1.03, 1.19), MBzP (HR: 1.13, 95% CI: 1.05, 1.22), MBP (HR: 1.11, 95% CI: 1.04, 1.19), and MCPP (HR: 1.06, 95% CI: 1.00, 1.12) showed an elevated risk of shortened gestational length per IQR change in their respective mean log-transformed concentrations. [...]summed DEHP metabolites had a nearly significant hazard ratio after accounting for the repeated measures of DEHP metabolites (HR: 1.07, 95% CI: 0.99, 1.16). [...]the accelerated failure time model in our analysis, which is typically used for right-skewed outcomes, is modeling a left-skewed outcome, namely gestational duration. [...]as a sensitivity check, we fit accelerated failure time models on a transformed version of gestational age, such that the log of the transformed gestational age was normally distributed (results not presented).